Special Issue in Electronics and Biomedical Engineering, Computer Science and Informatics
Research in Computing Science
Series Editorial Board Comité Editorial de la Serie
Editors-in-Chief: Editores en Jefe
Juan Humberto Sossa Azuela (Mexico) Gerhard Ritter (USA) Jean Serra (France) Ulises Cortés (Spain)
Associate Editors: Editores Asociados
Jesús Angulo (France) Jihad El-Sana (Israel) Jesús Figueroa (Mexico) Alexander Gelbukh (Russia) Ioannis Kakadiaris (USA) Serguei Levachkine (Russia) Petros Maragos (Greece) Julian Padget (UK) Mateo Valero (Spain)
Editorial Coordination: Coordinación Editorial
Blanca Miranda Valencia
Formatting: Formación
J. Humberto Sossa Azuela
Research in Computing Science es una publicación trimestral, de circulación internacional, editada por el Centro de Investigación en Computación del IPN, para dar a conocer los avances de investigación científica y desarrollo tecnológico de la comunidad científica internacional. Volumen 35, Mayo, 2008. Tiraje: 500 ejemplares. Certificado de Reserva de Derechos al Uso Exclusivo del Título No. 04-2004-062613250000-102, expedido por el Instituto Nacional de Derecho de Autor. Certificado de Licitud de Título No. 12897, Certificado de licitud de Contenido No. 10470, expedidos por la Comisión Calificadora de Publicaciones y Revistas Ilustradas. El contenido de los artículos es responsabilidad exclusiva de sus respectivos autores. Queda prohibida la reproducción total o parcial, por cualquier medio, sin el permiso expreso del editor, excepto para uso personal o de estudio haciendo cita explícita en la primera página de cada documento. Impreso en la Ciudad de México, en los Talleres Gráficos del IPN – Dirección de Publicaciones, Tres Guerras 27, Centro Histórico, México, D.F. Distribuida por el Centro de Investigación en Computación, Av. Juan de Dios Bátiz S/N, Esq. Av. Miguel Othón de Mendizábal, Col. Nueva Industrial Vallejo, C.P. 07738, México, D.F. Tel. 57 29 60 00, ext. 56571. Editor Responsable: Juan Humberto Sossa Azuela, RFC SOAJ560723 Research in Computing Science is published by the Center for Computing Research of IPN. Volume 35, May, 2008. Printing 500. Authors are responsible for the contents of their papers. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior permission of Centre for Computing Research. Printed in Mexico City, May, 2008, in the IPN Graphic Workshop – Publication Office.
Volume 35 Volumen 35
Special Issue in Electronics and Biomedical Engineering, Computer Science and Informatics Volume Editors: Editores del Volumen
Erik V. Cuevas-Jiménez Marco A. Pérez-Cisneros Daniel Zaldívar-Navarro Juan Humberto Sossa-Azuela Raúl Rojas
Instituto Politécnico Nacional Centro de Investigación en Computación
México 2008
ISSN: 1870-4069 Copyright © 2008 Instituto Politécnico Nacional Copyright © 2008 Instituto Politécnico Nacional Instituto Politécnico Nacional (IPN) Centro de Investigación en Computación (CIC) Av. Juan de Dios Bátiz s/n esq. M. Othón de Mendizábal Unidad Profesional “Adolfo López Mateos”, Zacatenco 07738, México D.F., México http://www.ipn.mxhttp://www.cic.ipn.mx Indexada en LATINDEX Indexed in LATINDEX Printing: 500 Impresiones: 500
Printed in Mexico Impreso en México
Preface
The use of engineering related subjects to study the systems of nature is one of the most significant trends of the century. Driven by the need for more precise scientific understanding, several fields have shown a massive growth expanding their boundaries in a quest to solve new challenging problems, generating new scientific knowledge and technology applications. This special issue presents original research papers on several engineering field of remarkable importance such as Electronics and Biomedical Engineering, Computer Science and Informatics. The volume is organized into four sections as follows: Control, Robotics and Artificial Intelligence, Signal and Image and Processing, Computer Science and Embedded Architectures and Biomedical Engineering The overall issue covers 21 papers carefully chosen from a peer-to-peer reviewing process on 47 contributions. Each submission was reviewed by at least three independent members of the Editorial Board and the final acceptation rate was 45.9%. As we are deeply thankful to all people involve in the preparation of this volume, it is to all the authors and the excellence of their research work that the true value of this book is owed. We also want to express our gratitude to all members of the editorial board of the volume and additional referees. The submission, reviewing and selection process was supported for free by the EasyChair© system. Erik Cuevas Marco A. Perez Daniel Zaldívar Humberto Sossa Raul Rojas May 2008
Table of Contents Índice
Page/Pág.
Control, Robotics and Artificial Intelligence Hardware Implementation of an Optimal Pole Placement Controller for a Liquid Level System……………………………….
Basil M. Al-Hadithi, Juan Suardíaz Muro, Susana Ortega Cisneros, Juan J. Raygoza Panduro, Juan A. López Riquelme
3
Discrete Time Nonlinear Identification via Recurrent High Order Neural Networks…………………………………………….
Alm Y. Alanis, Edgar N. Sanchez, and Alexander G. Loukianov
11
A Framework for Teleoperators Control…………………………... Emmanuel Nuño, Adolfo Rodriguez, Leopoldo Palomo, and Luis Basañez
21
Using Reinforcement Learning in Chess Engines………………… Marco Block, Maro Bader, Ernesto Tapia, Marte Ramirez, Ketill Gunnarsson, Eric Cuevas, Daniel Zaldivar, and Raul Rojas
31
Genetic Algorithm, a Multivariable and Multiobjective Approach…………………………………………………………...
Edgar Chavolla, Erik Cuevas, Daniel Zaldivar, Marco Perez and Alberto De La Mora
41
No Linear Haptic Rendering of Deformation and Cutting Based on Orthogonal Decomposition……………………………...
Gabriel Sepulveda, Vicente Parra, Omar Dominguez
51
Robust Synchronization of a Class of Robot Manipulators……….. Manfred Giljum and Gualberto Solis-Perales
63
Signal and Image and Processing Control of a Nurse Robot Using Voice Commands and Associative Memories……………………………………………..
Roberto A. Vazquez and Humberto Sossa
77
Detecting Scale-Sensitivity in Image Hierarchies for Coding and Compression………………………………………….
J. Alejandro Butron Guillen and Richard Harvey
87
Smart Camera Desing……………………………………………... Ivan Olaf Hernandez, Miguel Enrique Bravo Zanoguera, Guillermo Galaviz Yañes
97
Object Recognition Using Coupled Filters………………………... Jorge Hernandez Constante, Josue Alvarez Borrego, and Marco A. Cedano Olvera
107
Mixed Analog-Digital Implementation of the Semidiscrete Wavelet Transform………………………………………………...
Marco A. Gurrola Navarro and Guillermo Espinosa Flores Verdad
117
Computer Science and Embedded Architecture Improving Search and Publish of Knowledge by Means of Ontology in a Virtual Learning Enviroment………………………. Hector Diez Rodriguez and Jose Oscar Olmedo
Aguirre
127
Construction of an Optimal Solution for a Real World Routing Scheduling Loading Problem…….....................................
Jose F. Delgado, Laura Cruz Reyes, Juan J. Gonzalez, Hector Fraire H. and Rodolfo A. Pazos R.
137
Efficient Pattern Recalling Using Parallel Alpha-Beta Associative Memories……………………………………………. Mario Aldape Perez, Cornelio Yanez Marquez
and Oscar Camacho Nieto
147
Synchronization of Complex Networks withNonidentical Nodes……………………………………………………………… Gualberto Solis Perales and Daniela
Valle Rodriguez
157
Hot Rolling Scheduling Optimization Problem…………………… Carlos A. Hernandez Carrion, Hector Fraire Huacuja Karla Espriella Fernandez, Guadalupe
Castilla Valdez, and Juana Mancilla Tolama
165
Biomedical Engineering Analysis of 5 Source Separation Algorithms on Simulated EEG Signals……………………………………………………….. Ricardo Salido Ruiz, Rebeca Romo Vazquez, Radu Ranta and Lorenzo Leija
177
SISELS: A Mediation System for Giving Access to Biology Resources........…………………………………………
Gabriela Montiel Moreno, Jose Luis Zechinelli Martini,and Genoveva Vargas Solar
187
Fuzzy Gain Scheduling of PI Controller for an Anaerobic Digester……………………………………………………………. Albino Martinez Sibaja, Ruben Posada Gomez,
Alejandro Alvarado Lassman, Manuel Adam Medina and Carlos Astorga Zaragoza
199
Response of the Gravity-Inertial Mechanoreceptors During a Fall: Mathematical Model………………………………...
Vladimir Aleksandrov, Tamara Aleksandrova, Rosario Vega, Gregorio Castillo, Maribel Reyes, Yaneri Aguilar Aida Ortega, Nelly Shulenina and Enrique Soto
209
Biomedical Engineering
Response of the gravito-inertial
mechanoreceptors during a fall: a mathematical
model
Vladimir Aleksandrov1,2, Tamara Alexandrova1,2
, Rosario Vega3,
Gregorio Castillo2, Maribel Reyes2, Yaneri Aguilar3, Aída Ortega
3,
Nelly Shulenina1, and Enrique Soto3
1 Moscow State University, Leninskie Gory, 119991, Moscow, Russia
[email protected] de Ciencias Físico Matemáticas, Universidad
Autónoma de Puebla, Apartado Postal 1152, Puebla, Pue. C. P. 72000,
México
[email protected] Instituto de Fisiología, Universidad Autónoma de Puebla, Apartado Postal
406, Puebla, Pue. C. P. 72570, México,
Abstract. Various types of vestibular prosthesis prototypes have been devel-
oped as an aid for treatment of equilibrium disturbances. One of the primary
tasks for improving these prosthetic devices is the development of output
stimulating impulses that may resemble the natural response of the vestibular
system. In this work, a mathematical model of the information output from the
gravito-inertial mechanoreceptor of the vestibular apparatus is presented. For
this, we have considered five compartments: mechano-electrical transduction,
adaptation of transduction, hair-cell ionic current, synaptic transmission, and
afferent neuron discharge. The numerical parameters of the model were ob-
tained from experiments that were done in the inner ear of the rat. The results
of the numerical analysis of the model showed that the mathematical modelling
may be used to construct an encoder system for the artificial sensors (micro-
accelerometer) contributing to the development of a reliable vestibular prosthe-
sis prototype.
1 Introduction
The vestibular system, as well as other sensory organs, is a complex structure in
which optimization of incident energy to impinge and stimulate specific sensory cells
takes place. In the vestibule, the semicircular canals and the otolithic organs allow the
perception of the influence of gravity and of inertial forces produced by changes of
the head position to provide information used to stabilize the gaze and the posture.
Receptor hair cells of the vestibular system convert the energy of a mechanical
stimulus and transmit information about it to the first afferent neurons and then to the
central nervous system. The functional scheme of the vestibular mechanoreceptor is
© E. V. Cuevas, M. A. Perez, D. Zaldivar, H. Sossa, R. Rojas (Eds.) Special Issue in Electronics and Biomedical Informatics, Computer Science and Informatics Research in Computing Science 35, 2008, pp. 209-218
(Paper received on February 26, 2008, accepted on April 15, 2008)
shown in Figure 1. Displacement of the sensory hair bundle activates the transduction
process that originates a transducer ionic current. This leads to potential change in the
cell membrane that activates various voltage- dependent ionic channels in the hair
cell. This series of events finally produces a voltage- dependent activation of calcium
channels, and the subsequent activation of the neurotransmitter release machinery
leading to synaptic activation of the afferent neurons in the vestibular nerve. The
primary afferent neurons integrate the activity from various synaptic sources and
accordingly generate a series of action potentials. These afferent impulses are the
output from the vestibular mechanoreceptor. In the scheme (Figure 2) two levels of
control are taken into account: intrinsic based on the mechanism of adaptation of the
transducer current and extrinsic based on the operation of the efferent innervation.
In this work, we present a compartmental model of the vestibular mechanorecep-
tor in which we have considered five compartments: mechano-electrical transduction,
adaptation of the transduction mechanism, hair cell ionic currents, synaptic transmis-
sion, and afferent neuron discharge. In relation to the control mechanism, we only
considered the adaptation of the transducer mechanism. In this paper we consider the
union of the mathematical models that were presented earlier as a gravito-inertial
mechanoreceptor mathematical model.
2 Mathematical Model
Let us examine the extreme situation: initial stage of the uncontrolled fall of a man in
the sagital plane (during 100 ms), when there is still a possibility of the vertical pose
stabilization. As shown [2], the greatest reaction of hair cells to the mechanical
stimulus, which leads to the fall, occurs for the cells situated along the axis of the
sensitivity of the macula of sacculus, orthogonal at the local vertical line at the initial
moment of the fall (Fig. 1).
The sacculus, just as the utriculus, is a multi-dimensional accelerometer that makes
it possible to obtain information about the apparent acceleration of the otolith mem-
brane from many directions of sensitivity. Only one of these directions interests us, as
mentioned above. In connection with this, we will not consider the mathematical
model of the dynamics of the whole otolith membrane on the plane that is parallel to
the plane of the macula, and the response to this stimulus of many hair cells and pri-
mary afferent neurons, but only the dynamics along the axis of sensitivity that was
determined above. Hair cells located along the considered axis of sensitivity (Fig. 1),
in which the positive direction coincides with the direction of the forward fall (they
are located before the striola – reversal line), and the hair cells in which the positive
direction coincides with the direction of the backward fall (they are located after the
striola), we will consider the reactions of two hair cells with opposite polarity. We
have designated the hair cell altogether with the primary afferent neuron as vestibular
mechanoreceptor.
The term “gravito-inertial mechanoreceptor” for the stabilization of the vertical
210 Vladimir Aleksandrov et al.
Fig. 1. Functional polarity of the sacculus hair cells.The striola is designated by the dotted
line. Arrows represent the direction of maximal sensitivity for hair cells (Spoendlin H.H. In:
Wolfson R.J.,ed. The vestibular system and its diseases. Philadelphia, 1966, University of
Pennsylvania Press)
position will be used to name the set of the three mathematical models: the first of
them describes the dynamics of the displacement (xs) [3] of the otolith membrane
along the axis of sensitivity that is being considered, and the other two models de-
scribe the response of the mechanoreceptors of the opposite directions sensitivity
(x=± xs) of the otolith membrane (Fig. 2).
In sections 2.1 and 2.2 the basic model consisting of the “Current dynamics in hair
cells” and “Afferent neuron dynamics” are presented. In section 2.3 these two blocks
are connected by the “Synaptic transmission” block. Also described are the input of
the “Mechanoelectrical transduction” and the “Transducer adaptation” to prolonged
mechanical stimulus.
Fig. 2. Scheme of the vestibular mechanoreceptor compartments considered in the model.
2.1 Current dynamics in hair cells
The model is based on the Hodgkin-Huxley equations. This is a simplified model,
assuming that the dynamics of a hair cell may be described using a single total ionic
current IT [1], where IT is the sum of the principal currents of the hair cells. The model
is summarized in (1).
Response of the Gravity-Inertial Mechanoreceptors … 211
,11
1 LTTrm IIIdt
dVC −−−= ),)(( 121
3
TTT EVhhmgI −+= ,111 VgI LL =
,)()( 11 mVmdt
dmV STm −=τ
,)()( 111
1
11hVhq
dt
dhV STh −=τ (1)
,)()( 212
2
12hVhq
dt
dhV STh −=τ
Here IT is the total ionic current; m is the parameter that specifies the current acti-
vation process; h is the parameter that specifies the current inactivation process; gT is
the maximum conductance; IL is the leakage current; and Icom is, under natural condi-
tions, the current flowing into a hair cell through the transduction channels (Icom = -
ITr), or in the experiments, the command current. The inactivation parameter h has
two constituents (h=h1+h2) corresponding to the potassium channels with fast and
slow inactivation time constants. Functional parameters are shown in table 1, where
mmin, hmin, τmin, τmax, Vac, Vr, Vh, Sac, Sr and Sh are coefficients of sigmoidal fitting
curves that containing this functional parameters; kh1, kh2, bh1, bh2 are coefficients of
approximation for the fast and slow inactivation time constants.
Table 1. Functional parameters of the model used in (1)
Name Functional Form Name Functional Form
Steady-state Acti-
vation
−−+
−+=
ac
ac
ST
S
VV
mmVm
)(exp1
1)(
1
minmin1
Fast
Inactivation
Time Constant
11111 )( hhh bVkV +=τ
Activation Time
Constant
−+
−+=
τ
τ
ττττ
S
VVVm
1
minmaxmin1
exp1
)(Slow
Inactivation
Time Constant
21212 )( hhh bVkV +=τ
Steady-state
Inactivation
−+
−+=
h
h
ST
S
VV
hhVh
1
minmin1
exp1
1)(
Table 2. Hair cell numerical parameters.
Parameter Semicircular
Canal
Parameter Semicircular
Canal
Parameter Semicircular
Canal
Cml 11.26 pF Sτ 15.68 mV Vh -9.82 mV
gL 2.32 nS Vac -25.36 mV Sh 21.96 mV
gT 77.84 nS Sac 15.06 mV hmin 0.73
ET -79 mV mmin 0.37 r 3
Icom 0 pA kh1 0.82 ms/mV q1 1/2
τmax 77.58 ms kh2 1.26 ms/mV q2 1/2
τmin 6.55 ms bh1 55.86 ms
Vτ -52.23 mV bh2 282.38 ms
212 Vladimir Aleksandrov et al.
The numerical parameters of the model were obtained from experimental voltage-
clamp recordings of the isolated hair cells from the semicircular canal of the rat [5,6]
(Table 2).
Fig. 3. Voltage response trajectories obtained for Icom = 0. These traces were obtained for
different initial conditions taken for system (1): (1) V0 = -57.67 mV, m = 0.0041, h1= 0.9,
h2=0.1; (2) V0 = -52 mV, m = 0.240, h1= 0.8, h2 = 0.2; (3) V0 = -57.67 mV, m = 0.340, h1=
0.8, h2 = 0.2; (4) V0 = -57.67 mV, m = 0.440, h1= 0.8, h2 = 0.2
In figure 3, there are the voltage response trajectories obtained for Icom=0. The
model predicted a resting potential of -57 mV obtained with the values in table 1. The
dynamics of the hair cell membrane potential, obtained with the use of the mathe-
matical model shown in (1), qualitatively coincides with the results of the physiologi-
cal experiments (Fig. 4).
Fig. 4. Traces showing a typical voltage response of a hair cell obtained from the rat’s
semicircular canals subjected to current pulse injection (from -0.1 to 0.5 nA) (the dotted line
shows the zero voltage)
Response of the Gravity-Inertial Mechanoreceptors … 213
2.2 Afferent neuron dynamics
In describing the activity of the primary afferent bipolar neuron, a Hodgkin-Huxley-
type model was also used. The parameters were calculated using experimental results
obtained from cultured vestibular afferent neurons of the rat [7,8]. The use of rat
parameters in this model is the first modification with respect to the original Hodg-
kin-Huxley model. Other two modifications were about the time constants: an inacti-
vation parameter for outward current “hK” other is a modification in the mathematical
model original of Hodgkin-Huxley where h + n =0.8, here we have next modification
h + n=C(V2), C (V2) is an experimental results. The right part of this equality has a
constant value for each V2. In addition, our model have a complex description for
potassium current )( 2
4max
KKKK VVhngI −= . Based on these modifications and assuming
that τm = 0 and τhk = constant an intersection of two isoclines as an unstable point of
repose was found. Therefore, a limit cycle and the correspondent auto-oscillations
were also found.
Fig. 5. Isoclines of the simplified and of the modified Hodgkin-Huxley model
The modified and simplified Hodgkin-Huxley model (see above) for the action
potential generation takes the form as shown in [4].
max max 3 max 422 1 2 2 2 2 2( ) ( ) ( )( ( ) )( ) ( )m com L L Na Na K k K
dVC I V g V V g m V C V n V V g n h V V
dt ∞= − − − − − − −
nVndt
dnVn −= ∞ )()( 22τ (2)
)()()( 222 VhVhdt
dhV kk
khk −= ∞τ
The coefficients max
Nag , max
Kg , max
Lg belong to confidence intervals in accordance
with the experimental results. Table 4 present their values which correspond to the
214 Vladimir Aleksandrov et al.
greatest interval between two points (I1, I2) of the bifurcation of Hopf [9]. These
points indicate the appearance and disappearance of the auto-oscillations.
Table 3. Parameters of the model for vestibular afferent neurons (2)
Activation stable state gNa
+−+
=∞
2.5
)8.33(exp1
1)(
2
2V
Vm
Inactivation stable state gNa
++
=∞
9.9
5.60exp1
1)(
2
2V
VhNa
Activation stable state gK
+−
+=∞
5
)35(exp1
1)(
2
2V
Vn
Inactivation time constant gNa
5.0
5
30exp
15
79exp01.0
1)(
22
2 +
++
−
++
=VV
VhNaτ
Activation time constant gK
++
−
+=
20
30exp
15
25exp
68)(
22
2VV
Vnτ
Inactivation stable state gK
7329.0
24986.10
87968.33exp1
7329.096408.0)(
2
2 +
++
−=∞
VVhK
Inactivation time constant gK
500
10
25exp
15
15exp
1250)(
22
2 +
++
−+
=VV
VKh
τ
Table 4. Numerical parameters of the model (2).
Constants Units Chosen Value Constants Units Chosen Value
Cm2 µF/cm2
1 max
Nag mS/cm2
2.3
VNa mV 52 max
Kg mS/cm2
2.4
VK mV -84 max
Lg mS/cm2
0.03
VL mV -63 Icom µA/cm2
1 to 150
The amplitude of the auto-oscillations depends on the value of Icom (where Icom= -
Isyn). The first point of bifurcation I1 = 0.6 µA/cm2, the second point of bifurcation of
Hopf I2 = 165.3 µA/cm2.
2.3 Synaptic transmission, mechano-electrical transduction and transducer
adaptation
Data from experimental studies of synaptic transmission in the bullfrog inner ear [10]
were used for the association of the blocks that describe the dynamics of ionic cur-
rents in the hair cell and in the primary afferent neuron (Fig 2). The curve shown in
figure 6 shows the relationship between the voltage in the hair cell (V1 in model 1)
and the synaptic current in the afferent neuron (Isyn equivalent to Icom). The maximum
synaptic current was hypothesized to be equivalent to 40 µA/cm2.
Let us add, to the chain of three blocks just described, an input block for the me-
chano-electrical transduction mechanism [1], and the mechanism of the hair cell
Response of the Gravity-Inertial Mechanoreceptors … 215
transducer adaptation to the prolonged mechanical stimuli. The mathematical model
of these two mechanisms is represented in the form of equation (3).
The adaptation mechanism is given by [4]. Where s is the adaptation parameter; τad
is a time constant; k is a gain constant; ITr is the transduction current; ITr0 is the trans-
duction current in stationary state; p(x,s) is the probability of the opening of the canal;
x is the displacement of a hair bundle.
1( ); ( , )( ); ( , );ad Tr Tro Tr Tr Tr Tr Trs s k I I I g x s V E g g p x sτ + = − = − =&
;100;4.1 msnSg adTr == τ0
1
1( , ) ;
1
p x sx s x
Exps
= + −
+ −
(3)
0;3.0;pA 4.14I ;03.0 ; 2.0 0Tr01 ==−=== TrEmxkms µµ
Using this system, a mathematical model of the vestibular mechanoreceptor infor-
mation output was obtained. It consists of equations (1), (2), (3), tables 1, 2, 3, 4 and
the graph in Figure 6.
Fig. 6. Relationship between the membrane voltage in the hair cell and the synaptic
current in the afferent neuron. Continuous line represent the best fit to the experi-
mental data [10].
3 Numerical Results
After the association of all blocks of the model and the analysis of the dynamics of
ion currents in the hair cell and in primary afferent neuron, the numerical parameters
were selected on the basis of physiological experiments (tables 1, 2, 3 and 4). The
results of the calculations for the initial stage of the fall are shown in Fig 7 (the incli-
nation forward to 30 degrees-Figure 7A-). Development of membrane potential V1 of
hair cell as a result of the mechanism of adaptation acting against the background of
216 Vladimir Aleksandrov et al.
the mechanical stimulus are shown in Figure 7 B. Finally we have a secondary infor-
mation in the form of afferent impulses of the primary neuron with frequencies vary-
ing from 20 Hz to 40 Hz.
Fig. 7. The process of the information output from the vestibular mechanoreceptor. In A,
mechanical stimulus displaces a hair cell bundle. Stimulus is absent during the first 100 ms
(stationary situation of rest); in the course of the following 200 ms the hair bundle is displaced
1 µm. In B, the voltage response of the hair cell reflects the activation of the transducer adapta-
tion mechanism. In C, the output of the model in the form of action potentials in the primary
afferent neuron.
4 Conclusion
The numerical results indicate that the mathematical model of information processing
in the gravito-inertial mechanoreceptor resembles the activity of the natural sensor as
studied experimentally.
In the development of vestibular prosthetic devices a transfer function derived
from the recordings in the monkey inner ear [11] has been used to convert the analog
output of the device to a pulse train useful to stimulate the afferent nerve. We propose
that the use of more realistic models based on the physiological knowledge and using
parameters from animal experiments, will endow prosthetic devices with greater
coding capabilities than those of devices using simpler transfer functions.
A
B
C
Response of the Gravity-Inertial Mechanoreceptors … 217
Our results demonstrate that the development of an integrated mathematical model
of the function of vestibular endorgans is feasible and that it will resemble vestibular
system coding capabilities.
It is concluded that the proposed mathematical model may be used to construct an
encoder system for artificial sensors (eg: microaccelerometer and microgyroscope)
contributing to the development of a reliable vestibular prosthesis prototype.
Acknowledgements
This work was performed with the support of UC-MEXUS-CONACyT grant from the
University of California at Riverside project to ES; Russian state contract No
02.512.11.2161 for studies with the collaboration of foreign scientific institutions to
VA: and VIEP-BUAP grant No. 0054 to VA.
References
1. Alexandrov, V.V., Almanza, A., Kulikovskaya, N.V., Vega R., Alexandrova, T.B.,
Shulenina, N.E., Limón, A. and Soto, E.: A Mathematical Model of the Total Current Dy-
namics in Hair Cells. In Sadovnichii, V.A.(ed): Mathematical Modeling of Complex Infor-
mation Processing Systems, Moscow State University,(2001) 26-41
2. Alexandrov, V.V., Alexandrova, T.B., Astakhova, T.G. Kulikovskaya, N.V., Kurilov, V.I.,
Migunov, S.S., Shulenina, N.E., Soto, E., Vega, R.: A mathematical Model of the Response
of Semicircular Canal and Otolith to Head Rotation under Gravity. J Gravitational Physiol-
ogy 11(2)(2004) 25-26
3. Alexandrov, V.V., Alexandrova, T.B., Migunov, S.S.: The Mathematical Model of the Gra-
vity-inertial Mechanoreceptor. Moscow University Mechanics Bulletin, No. 2, (2006) 59-64
(in russian)
4. Alexandrov, V.V., Mikhaleva, E.Yu., Soto, E., García Tamayo, R.: Modification of Hodg-
kin-Huxley Mathematical Model for the Primary Neuron of Vestibular Apparatus. Moscow
University Mechanics Bulletin, No. 5 (2006) 65-68 (in russian)
5. Almanza, A., Vega, R. and Soto, E.: Calcium Current in Type I Hair Cells Isolated from the
Semicircular Canal Crista Ampullaris of the Rat. Brain Research 994 (2003) 175-180
6. Almanza, A., Vega, R., Naverrete, F. and Soto, E.: Nitric oxide modulation of L type Ca2+
current in sensory hair cells. J Neurophysiology 97 (2007) 1188-1195
7. Soto, E., Limón, A., Ortega, A. and Vega, R.: Características Morfológicas y Electrofisi-
ológicas de las Neuronas del Ganglio Vestibular en Cultivo. Gaceta Médica de México 138
(2002) 1-14
8. Limón, A., Pérez, C., Vega, R. y Soto, E.: IK, Ca Current Density is Correlated with Soma
Size in Vestibular Primary Afferent Neurons. J Neurophysiology 94 (2005) 3751-3761
9. Troy, W.C.: The Bifurcation of Periodic Solutions in the Hodgkin-Huxley Equations. Quar-
terly of Applied Mathematics 36 (1978) 73-83
10. Keen, E.C. and Hudspeth, A.J.: Transfer Characteristic of the Hair Cell’s Afferent Synapse.
Proc. Natl. Acad. Sci. USA 103 (2006) 5437-5542
11. Fernández, G. and Goldberg, J. Physiology of Peripheral Neurons Innervating Otolith
Organs of the Squirrel Monkey. II. Directional Selectivity and Force-Response Relations. J
Neurophysiology 29 (1976) 985-995.
218 Vladimir Aleksandrov et al.
Author Index Índice de autores
Aguilar Yaneri 209 Alanis Alm 11 Aldape Mario 147 Aleksandrov V. 209 Aleksandrova T. 209 Alvarado A. 199 Alvarez Josue 107 Astorga Carlos 199 Bader Maro 31 Basañez Luis 21 Basil Hadithi 3 Block Marco 31 Bravo Miguel 97 Butron Alejandro 87 Camacho Oscar 147 Castilla Gpe. 165 Castillo Gregorio 209 Cedano Marco 107 Chavolla Edgar 41 Cruz Laura 137 Cuevas Erik 31,41 De la Mora A. 41 Delgado Jose 137 Diez Hector 127 Dominguez Omar 51 Espinosa G. 117 Espriella Karla 165 Fraire Hector 137,165 Galaviz G. 97 Giljum Manfred 63 Gonzalez Juan 137 Gunnarsson K. 31 Gurrola Marco 117
Harvey Richard
Harvey Richard 87 Hernandez Carlos 165 Hernandez Ivan 97 Hernandez Jorge 107 Leija Lorenzo 177 Loukianov A. 11
Mancilla Juana 165 Martinez Albino 199 Medina Manuel 199 Montiel Gabriela 187 Muro Juan 3 Nuño Emmanuel 21 Olmedo Jose 127 Ortega Aida 209 Ortega Susana 3 Palomo Leopoldo 21 Parra Vicente 51 Pazos Rodolfo 137 Perez Marco 41 Posada Ruben 199 Ramirez Marte 31 Ranta Radu 177 Raygoza Juan 3 Reyes Maribel 209 Riquelme Juan 3 Rodriguez A. 21 Rojas Raul 31 Romo Rebeca 177 Salido Ricardo 177 Sanchez Edgar 11 Sepulveda G. 51 Shulenina Nelly 209 Solis Gualberto 63,157 Sossa Humberto 77 Soto Enrique 209 Tapia Ernesto 31 Valle Daniela 157 Vargas Genoveva 187 Vazquez Roberto 77 Vega Rosario 209 Yanez Cornelio 147 Zaldivar Daniel 31,41 Zechinelli Jose 187
Editorial Board of the Volume Comité Editorial del volumen
Alejandra Santoyo
Alejandro Morales
Alma Alanis-Garcia
Antonio García
Basil M. Al-Hadithi
Carlos Lopez
Cuauhtémoc Aguilera
Didier WOLF
Dimitris Kalamatianos
Edgar N. Sanchez
Eduardo Ruiz
Efrén Aguilar
Enric Fossas
Ernesto Tapia
Felix Von Hundelshausen
Francisco Gómez
Freddy Lopez Villafuerte
Gustavo Sutter
Heriberto Zavala
Jagannathan Sarangapani
Jaime Alvarez
Jesus Palomino
Jochen Schiller
John Gray
José E. Rayas
Josúe Alvarez
Juan P. Garcia
Juan Raygoza
Juan Suardiaz
Luis Basañez
Mario Siller
Marco Block
Mark C. Readman
Marte Ortegon
Michael Himmelsbach
Michael Saliba
Mo Jamshidi
Nancy Arana-Daniel
Peter Cook
Peter Wellstead
Piotr Dudek
Radu Ranta
Raúl Rojas
Raúl Suárez
Richard Middleton
Rohan Munasinghe
Ronald Lasky
Susana H. González
Susana Ortega
Valérie LOUIS-DORR
Zoltan Kutalik
de la Dirección de Publicaciones del Instituto Politécnico Nacional
Tresguerras 27, Centro Histórico, México, D.F. Mayo 2008.
Printing 500 / Edición 500 ejemplares.
Impreso en los Talleres Gráficos
© E. V. Cuevas, M. A. Perez, D. Zaldivar, H. Sossa, R. Rojas (Eds.) Special Issue in Electronics and Biomedical Informatics, Computer Science and Informatics Research in Computing Science 35, 2008, pp. 87-96
Detecting Scale-Sensitivity in Image Hierarchies for Coding and Compression
J. Alejandro Butrón Guillén1 and Richard Harvey2
1Centro Nacional de Actualización Docente CNAD Estanislao Ramírez s/n, col. Selene. México 13420, D.F.
2School of Computing Sciences, University of East Anglia Norwich, NR4 7TJ, UK.
Contact: [email protected], [email protected]
Abstract. This paper examines human sensitivity to errors introduced by two types of lossy image coders. Our interest is a special type of morphological scale-space tree called a sieve because such trees are thought to be useful for image understanding and other purposes, but we also examine a conventional lossy coder: JPEG. The paper introduces a new way of measuring image quality, a type of Turing test and we show how the method can be normalized to compare different images and coders. We conclude that content of the image can have a significant effect on the perception of image quality.
1 Introduction
Motivated by MPEG-7 and MPEG-4, image and video analysis and compression based on connected-set mathematical morphology ([1] for example) have become topics of some interest. Lossy compression based on such techniques typicall y involves deleting small -scale regions and/or coding regions using an approximation to their true shape. A key property of many connected set methods is their hierarchical structure [1], [2], in which the image is represented as a tree with small-scale regions as the leaves and root representing the whole image. Such trees can become large, it is therefore of some interest to know how many of the leaves may be deleted from a tree without affect the quali ty perceived by a human observer. The sensiti vity of observers to the small scale contained in images is thus investigated and reported. Images are fil tered with the sieve, a type of hierarchical connected-set representation, and a subjective test devised that allows us to measure the effect of deleting regions from the hierarchy.
The rest of the paper is organized as follows. In Section 2, we review the sieve
algorithm and describe the hierarchical structure for decomposing images into hierarchies of contours. Section 3 presents the experimental setup for detecting the sensiti vity to contours, together with a discussion of the results, and in Section 4 we make some initial conclusions for the development of a compressor based on the sieve.
Top Related